miladalsh/spatial_traveres_results_docent-val-Qwen3-VL-8B-Instruct-random_no_ref

该数据集包含一个验证集，共有114个示例，总大小约为191MB。每个示例包含以下特征：唯一标识符（uuid）、步骤编号（step）、项目列表（items，字符串类型）、分段ID列表（segment_ids，整数类型）、分段位置列表（segment_positions，浮点数列表的列表）、详细参考描述（detailed_reference_description，字符串类型）、生成描述（generated_description，字符串类型）、提示词（prompt，字符串类型）以及标注图像（annotated_image，图像类型）。数据集可能用于图像标注、描述生成或视觉语言任务，涉及对图像内容进行文本描述和分段处理。

This dataset includes a validation set with 114 examples and a total size of approximately 191MB. Each example contains the following features: a unique identifier (uuid), a step number (step), a list of items (items, string type), a list of segment IDs (segment_ids, integer type), a list of segment positions (segment_positions, list of float64 lists), a detailed reference description (detailed_reference_description, string type), a generated description (generated_description, string type), a prompt (prompt, string type), and an annotated image (annotated_image, image type). The dataset is likely designed for image annotation, description generation, or vision-language tasks, involving textual descriptions and segmentation of image content.